LAUSR

Abstract The phenomenon of stochastic resonance is studied in an asymmetric tristable model driven by a periodic forcing and correlated multiplicative non-Gaussian noise and additive Gaussian white noise. Analytical expressions for the spectral amplification, together with mean first-passage time and information entropy production, are derived in the adiabatic limit. The theoretical and numerical results suggest that the phenomena of noise-enhanced stability, stochastic resonance as well as a double stochastic resonance are found by considering the effects of correlated noises and asymmetry of potential. Besides, the appropriate choice of asymmetry constant and cross-correlation strength can improve the response of the tristable system. The deviation of non-Gaussian noise from Gaussian character has a significant influence on the mean first-passage time and spectral amplification. Conditions under which the response of information entropy production to the periodic forcing can be optimized via the SR mechanism are identified. Finally, the proposed SR theory is applied to the bearing inner ring fault diagnosis by a general scale transformation. Specifically, the performance of fault detection can be improved effectively by exploiting the feature of asymmetry.